Using Learning Analytics within an e-Assessment Platform for a
TransFormative Evaluation in Bilingual Contexts
Samira ElAtia
1,*
and Eivenlour David
2
1
Division of Education, The University of Alberta, Edmonton, Alberta, Canada
2
Department of Computing Sciences, Faculty of Sciences, The University of Alberta, Canada
Keywords: Longitudinal Bilingual Evaluation, TransFormative Assessment, Learner’s Engagement, Learning
Analytics.
Abstract: Learning Analytics (LA) has the potential to be used as a unique and viable learning, teaching and research
tool to analyze data from longitudinal assessment. The online language assessment platform, Profil
Linguistique, is an innovative and useful tool, in that (1) it adapts to students learning abilities and progress
and gives them the chance to monitor their progress, (2) it uses data mining to provide reports to teachers
and administration who subsequently adapt the general language program in a Canadian university. From a
theoretical point of view, the testing construct identified as the basis of this online assessment tool would
engage students in progressing in their language competence in parallel with the courses they are taking. It
is a provocative and unique way to integrate and look at assessment as a teaching tool by using LA.
1 INTRODUCTION
Educational applications of data mining and learning
analytics are a growing trend in higher education
due to the vast amounts of data becoming available
from the growing number of uses within e-learning
environments (ElAtia and Ipperciel 2015; Zaiane
and Yacef 2015; Bakhshinategh et al. 2018).
Learning analytics (LA) is data analytics in the
context of learning and education. It involves
collecting data about learners ‘activities and
behaviours, as well as educational environments and
contexts, and using statistics and data mining
techniques to extract relevant patterns that reveal
how learning took place. According to Diaz and
Brown (2012), LA is the “use of data, statistical
analysis, and explanatory and predictive models to
gain insights and act on complex issues (…) about
the learners (3).” LA can be used to report measures
and patterns related to learning activities, or to
optimize learning activities and strategies and/or
learning environments. Two types of data can be
used for implementing LA. First, data generated by
hte learners themselves, and often referred to as
digital footprints. This type of data would enable us
*Dr. ElAtia is the principal investigator on this project and
is the contact person.
to implement techniques to carry out data mining
analyses leading to a holistic understanding of
students’ behaviour. Second, data supplied by
learners in the form of surveys and other
demographic and background information. This data
provides a foundation for building an information
system about students. Both types of data are
necessary to learn about how learners react, behave,
interact and use a specific e-learning environment.
LA also provides (1) insights on how effective are
such environments and (2) feedback for both future
improvement and potential wider use. In this
research project, we are using LA to empirically
study students data collected from an on-line
interactive assessment framework that targets
language competence. The objectives of LA can be
to report measures and patterns related to learning
activities, or to optimize learning activities and
strategies and/or learning environments.
For the last two years, a research team at the
University of Alberta worked on developing an
assessment model within an e-learning environment
that comprise of individualized diagnostic, formative
and summative evaluation modules using Moodle”
as an e-delivery platform and the Canadian
Language Benchmarks as competence indicators.
This assessment model, named ‘profil linguistique
(PL)’ provides a learner-centered and an integrated
674
ElAtia, S. and David, E.
Using Learning Analytics within an e-Assessment Platform for a TransFormative Evaluation in Bilingual Contexts.
DOI: 10.5220/0007860406740680
In Proceedings of the 11th International Conference on Computer Supported Education (CSEDU 2019), pages 674-680
ISBN: 978-989-758-367-4
Copyright
c
2019 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
approach to students own evaluation of language
achievement and is meant to supplement content-
based courses. The PL allows us to offer a language
program anchored in a continuous contact with the
language (Berman and ElAtia 2008) within content-
based courses (ElAtia 2011, ElAtia and Lemaire
2013) and which takes into consideration learners
strategies, motivation, constructive feedback and
self-engagement within the theories of language
acquisition, (Cohen 2012, 2011, 2010). The premise
is that by taking their language learning in-hand in
the PL, students will have a holistic idea of their
progress and most importantly would identify their
linguistic weaknesses and elaborate personal
strategies for linguistic improvement. From a
research perspective, data produced and collected
from the PL constitutes a solid foundation for
carrying LA studies that will bring new insights into
understanding how students interact with various
learning materials within bilingual e-learning
environments. We hope to improve our
understanding of the strategies used by students to
solve problems, to correct mistakes and to
circumvent difficulties. Within an LA framework,
the objectives of this research are to:
(a) Understand what helps students succeed when
using this independent e-learning model; how to
use analytics to improve the design and
applicability of each parts in the PL in such
ways as to promote student success in engaging
in their own language learning, be it French or
English;
(b) Use data to understand the various subgroups
within the students’ population; to build a
comprehensive behavioral pattern for practical
direct input into regular classes.
(c) Develop specific data mining techniques to
extract, sort, store, cluster and analyze data,
determine patterns to seek from the data, and
maximize what is available to support students,
teachers and administration.
2 BILINGUAL ASSESSMENT AND
ENGAGMENT
2.1 Context of the Study
The Official Languages Act went into effect in 1969,
giving French and English equal official status
within Canada. Bilingualism has since been
promoted not only at the federal level and the
working of government but also provincially
through the French immersion language programs.
However, outside of certain centers, namely Quebec
and New Brunswick, French is in a minority context
within an English-dominant environment. The
competence in French in the English speaking part
of Canada is heavily influenced by several factors as
shown in figure 1. The language identity plays a big
part in the language development of students across
Canada.
Figure 1: Elements Impacting Language in Canada.
While French Immersion programs are common
across Canada, students wanting to carry out their
higher education choose to do so amongst a very few
bilingual programs and institutions such as Glendon
College at the York University and Campus Saint-
Jean at the University of Alberta. However, within
these bilingual programs, mastering the two official
languages is a requirement for admission and
students are faced with the challenge of having to
work on their second /other language (either French
or English) in order to graduate with the required
language competency. Program administrators face a
challenge in this situation because there is little
space within the majority of a 4 year program to add
language courses. The profile linguistique (PL) was
conceived as a testing and teaching tool that can be
carried out by students and support staff outside of
class time, and monitored by teachers and
administrators to offer a language support
opportunity to students in parallel to their content-
based course work. Assessment, in all its facets from
diagnostic, formative and summative, is used to
motivate and engage the learners.
Students in these bilingual programs face few
challenges that we hope the PL address:
linguistic heterogeneity
heavy content base coursework
language competence plateau
This study took place in a large university in
Western Canada, within its bilingual campus,
located within an English dominant environment,
and consequently, the students are in a francophone
linguistic minority context within an historical
French minority. Upon acceptance to university,
students take an online French and English
Using Learning Analytics within an e-Assessment Platform for a TransFormative Evaluation in Bilingual Contexts
675
placement test in order to identify their proficiency
level and to place in the appropriate class. The
purpose of PL is going beyond a placement test: it
aims to create an online platform that would allow
for in individualized language learning experience
using LA outside of classrooms.
2.2 Assessment for Language Learning
Assessment is generally an applied research field
that interconnects (a) learning styles and
preferences, (b) teaching strategies, and (b)
evaluation techniques. Assessment can serve both
learning and teaching; in fact, assessment can be a
powerful tool where assessment becomes a platform
in service of learning. Formative assessment can be
used as a tool to advance learning where students
take their learning into their own hands. Students’
performance exceeds expectations once they are
engage.
Figure 2: TransFormative Triangulation.
The PL is anchored in a transformative assessment
framework (Popham 2007), where the emphasis in
on evidence-based assessment and understandings of
students’ language progress from three primary
stakeholders: students, instructors and program
administrators. The purpose is not only to do a
survey, gather information, and establish a starting
point about students’ progress, but also to create a
loop of constructive feedback and assessment that all
stakeholders can use to improve and change learning
patterns throughout a two-year cycle. Figure 2
summarises the theoretical basis for the PL: from
raising awareness, to motivating students do better,
to finally an engagement on their part to learn.
As a first step, it is rudimentary to establish a
baseline for learning; we are then positioned to
determine progress against this baseline and provide
evidence of development (Astin, 1985, 1991; Banta
and Palomba, 2015). Importantly, assessment can
thus become a platform in the service of learning.
This is captured in the diagnostic assessment phase,
which contains both a detailed survey and a
competence test. The second, and substantive phase,
is the formative part of the Pl, that contains 5
modules, each containing a series of assessment
tools. Formative evaluation can be used as a tool to
advance learning, with its focus on students taking
on responsibility for their own learning (ElAtia and
Lemaire 2013). Indeed, evidence supports that
students’ performance exceeds expectations once
they are engaged in their learning (Kuh 2009;
Berman and ElAtia 2008). Teachers, on the other
hand, can change their pedagogical approaches to
address students’ needs, based on the feedback they
receive from students (Popham 2007, 2011). During
a learning process, consciously being aware and
actively engaged in the progress of learning has
proved central to learners fully endorsing the
objectives and outcomes they are exposed to in a
course (Harperand Quaye 2009; Schmidt 1990, Ellis
2008). Building on this principle of engaged
learning (versus more static, passive learning),
instructors and administrators clearly must draw
students’ attention to the details of the learning
process itselfand continual formative assessment
certainly aids this. The PL serves this purpose by
explicitly focusing on the student as an active part of
the learning process, it is a tool that can help
students awaken to awareness, thus facilitating
engagement. As students do the assessment, they
become aware of their progress; whether for
extrinsic or intrinsic purposes, they will be
motivated to improve their standing; and by doing
so, they will become more engaged and reflective of
their own learning, be it at the course level or at the
extracurricular one. Astin (1984, 1993) in
conjunction with Lawon et al (2012) and Lyster
(2010) insist on the effectiveness of pedagogical
techniques that learners would not otherwise use,
were they not centered on heightening awareness of
the learning process. In turn, this leads to motivation
to improve the implementation and learning the PL
provides an independent, non-intimidating, self-
monitoring platform that generates visual reports
(securely and confidentially) that users can build on
for a more engaged learning/teaching of languages
within an academic program. The assessment within
PL demarcates itself from other tools by allowing
users to (1) visually track progress, (2) pinpoint
acquired/not yet acquired attributes, and (3) build a
portfolio of artifacts as evidence of acquisition.
2.2.1 Canadian Language Benchmarks
Know in French as the Niveaux de compétences
langagiers canadiens are the national standards in
English and French for describing, measuring and
recognizing second language proficiency of adults in
Canada. They are founded on models of
A2E 2019 - Special Session on Analytics in Educational Environments
676
communicative competence; they are descriptive
scales of language ability both (ESL and FSL). The
Benchmarks are written on a 12 scale reference pints
along a continuum from basic to advanced
proficiency. These 12 benchmarks reflect
progression of the knowledge and skills that underlie
basic, intermediate and advanced ability among
adult ESL and FSL learners. Because they are a
Canadian products, that has been developed while
taking in consideration the cultural, social,
economic, political consideration in Canada, we
used them as our point of reference for developing
the content materials for the PL. The benchmarks are
also used by various employers in requiring special
French or English language competence. For the PL,
we created modules and various testing tools that
target level 5 to 9of the benchmarks; we did not take
into consideration level 10 to 12 because they relate
more to graduate education than to the
undergraduate one.
2.3 Description of the
Profil Linquisitique
This triangulation model was used during a year-
long formative assessment model where learning
analytics and educational data mining were
implemented to single our elements within students’
profiles that could predict their success. The online
platform of the PL was created in 2012 using
Moodle as the learning management systems; as
dynamic and computer adaptive platform with a data
collection model for process mining. The PL is a
bilingual comprehensive online assessment portal
that has three components as shown in figure 3
below:
1. The initial assessment phase contains (1) an
extensive survey about the language background of
the students and (2) a diagnostic test that serve to
identify the competence of the students both in
English and French.
2. The formative assessment phase contains five
modules calibrated with the Canadian Language
Benchmarks (CLB). Each module contains various
quizzes and tests that students can independently use
outside of class to target specific learning objectives
they want to improve for either French or English.
3. The achievement phase contains a high stakes
competency test of language for academic purposes
and for specific purposes such as business and
nursing.
Figure 3: Components of the Profil Linguistique.
3 ONLINE ASSESSMENT
PLATFORM
3.1 Learning Management System
The university where the study is undertaken, uses
Moodle as its learning platform. Moodle is designed
to provide educators, administrators and learners
with a single robust, secure and integrated system to
create personalised learning environments. It is an
open source product, meaning that the source code is
openly available and that there is an entire
community of people contributing to the
improvement of the project whether it be by fixing
bugs or creating new modules. eClass, powered by
Moodle, is the university’s centrally supported
learning management system. eClass houses the
university’s credit and non-credit courses. To access
these courses, users must (1) have a Campus
Computing ID (CCID) account and (2) be enrolled
in these courses. CCID accounts are created mainly
for university students within a few days after their
application. Figures 4, 5 and 6 show the progress of
the within the PL.
Even though PL is an assessment tool, it is
developed in eClass as a private non-credit course
and used to evaluate upcoming students. For the
purpose of generating data, we have made students
that are currently enrolled in language courses at the
university as PL’s participants for the duration of
this research. These students are enrolled and
grouped by their own cohorts at the course-level;
thus allowing teachers to monitor the progress of the
students and provide input and feedback on the
Using Learning Analytics within an e-Assessment Platform for a TransFormative Evaluation in Bilingual Contexts
677
content of the assessment modules. Upon entering
the course, students can choose between French and
English whichever language they are assigned to
be assessed in.
Figure 4: Gradual Progress in the PL.
Figure 5: Initial Log in screen.
Once in their learning space, students can choose the
language of their choice for the assessment. They
have the option to work on either language
depending on their preferences and language needs,
i.e. francophone tend to work on English more, and
the English speakers tend to work on their French
more.
Figure 6: The Diagnostic Assessment Phase.
The screenshots in figure 5 and 6 above describe the
first steps in the diagnostic assessment. The survey
is mandatory before the test. A part from the written
production that would need to be graded by teachers,
students, upon completing their assessment, receive
a table of their work broken down by item. The
administration gets a different aggregate report that
would allow it to advise students on future task to
undertake to improve language production.
4 INITIAL DATA RESULTS
To analyze data, we used (1) Python, a programming
language popularly used in data science, (2) Pandas,
a Python data analysis library, and (3) Microsoft
Excel to produce graphs. In eClass, we downloaded
the data from the survey and DA in CSV file format.
To generate reports, the participants are de-identified
and instead, have been assigned a random unique
ID.
Concurrently, we carried analyses using the
survey as our first point of reference to understand
how students navigate through the various
assessment modules in the PL, their answers and
feedback, to investigate clusters, identify outliers,
and map out rule associations that would shed light
on how the various groups and subgroups students
interact within e-assessment models.
The focus of this article is the initial phase.
Using the survey and the diagnostic assessment, we
were able to identify unique trends that would
explain students’ behaviour while interacting with
the language diagnostic tests.
The figures below show how we were able to
Figure 7: Overall Performance by Type of background
Education.
In Figure 7 above and Figure 8 below, we look at
the performance by type of prior education program
in the elementary and secondary level. The cluster
A2E 2019 - Special Session on Analytics in Educational Environments
678
analysis shows that the previous language of
instruction of students does in fact predict their
competence in various parts of the PL.
Figure 8: Partial Performance by Type of background
Education.
The type of education that students had before
joining this program has an impact on their language
competence. Their performance in the test clearly
can be clearly predicted by which program they
attended. Moreover, performance in particular parts
in the diagnostic test can we traced to their
background education. By doing do, the PL can help
us identify trends among certain groups; and
consequently, be able to adapt the teaching material
and to target specific component of language
acquisition that each group needs to work on, or
strengthen. Such reports will help the administration
identify where resources should be allocated in order
to help students in the program attain the adequate
degree of language competence.
In graphs 9 and 10, we delved into specific parts
of the diagnostic test. The purpose was to be able to
generate reports for individual examinees. Both
graphs show the performance broken by parts.
Students will be able to find out which of the
language component they master, and which one
still need some work. By comparing each student to
the mean, they can visualise their progress, become
aware of their current language status with the hope
to motivate them to do better and seek help from the
various resources that are available to them in their
programs. They can then pursue the formative
modules within the PL.
Overall, the initial results showed us various
clustering and trends that we were unsuspecting of.
It helps us revise the theory of immersion students
vs. the French as second language students.
Figure 9: Overall Performance within the 4 Parts of the
Diagnostic Test.
Figure 10: A Comparative Visual of one subject with the
Group.
5 CONCLUSIONS
We aim to build an LA model that would allow us to
carry in-house research that would benefit students,
the university and academic communities. This is
the first round of analysis that we carry on the PL.
We realise that one of the drawbacks is the small
pool of examinees. However, we are confident as
Using Learning Analytics within an e-Assessment Platform for a TransFormative Evaluation in Bilingual Contexts
679
more students are using the PL, we will accumulate
data that will help us understand behaviour,
performance and needs of our students.
This project will open new doors for improving
our understanding of students’ interaction within e-
learning models specific to language competence.
As a result, we will, longitudinally and
comprehensively, be able to explore data relating to
students. The development of a data warehouse
model and data mining algorithm for the LA model
will have utility for secondary and post-secondary
language programs across Canada. Using this
complete assessment framework and the ensuing LA
model, both the substantive and the qualitative
results, will allow program teachers and
administrators to holistically follow students’
progress throughout their academic careers and to
customize their programs for optimal success. Our
initial data and analysis does support our thesis and
we plan to carry out more extensive analysis as more
robust data is collected.
ACKNOWLEDGEMENTS
This research has been possible from the Provost
office from the McCalla Professorship, awarded to
Dr. ElAtia in 2017-2018
REFERENCES
Astin, A. (1985). Achieving Educational Excellence: A
critical Assessment of Priorities and Practices in
Higher Education. San Francisco, CA : Jossey-Bass.
Astin, A. and Antonio, A.L. (1993). Assessment for
Excellence: The Philosophy and Practice of
Assessment and Evaluation in Higher Education.
Lanham, MD: Rowman and Littlefield Publishers
Banta, T.W. and Palomba, C.A (2015). Assessment
Essentials: Planning, Implementing, and Improving
Assessment in Higher Education. San Fransisco. CA:
Jossey-Bass
Bakhshinategh, B., Zaiane, O.R., ElAtia, S. and Ipperciel,
D. (2017). Educational data mining applications and
tasks: A survey of the last 10 years. Education and
Information Technologies, 22(4), 117.
Berman , R and ElAtia, S. (2008). Classroom time as a
factor in instruction of English for academic purposes.
In Z. Ibrahim et S. Makhlouf (eds.). Linguistics in an
Age of Globalization. Cairo: AUC Press, 131-145.
Cohen, A.D. (2012). Strategies: The interface of styles,
strategies, and motivation on tasks. In S. Mercer, S.
Ryan, and M. Williams (Eds.), Language learning
psychology: Research, theory, and pedagogy (136-
150). Basingstoke, England: Palgrave Macmillan.
Cohen, A.D. (2011). Strategies in learning and using a
second language. Harlow, England: Longman Applied
Linguistics/Pearson Education.
Cohen, A.D. (2010). Focus on the language learner:
Styles, strategies and motivation. In N. Schmitt
(Ed.), An Introduction to Applied Linguistics (161-
178). 2nd ed. London: Hodder Education.
Diaz, V. and Brown, M. (2012). Educause Learning
Initiative: Learning Analytics. Retrieved from:
http://www.educause.edu/ELI124
ElAtia, S. (2011). Enhancing Foreign Language Writing in
Content-Based courses. Journal of Theory and
Practice in Education, 7, 2: 192-206.
ElAtia, S. and Ipperciel, D. (2015). At the Intersection of
Computer Sciences and Online Education:
Fundamental Consideration in MOOCs Education.
Educational Letter, 11 (2), 2 7.
ElAtia, S and E. Lemaire. (2013). Engager les apprenants
dans l’autocorrection de leurs productions écrites : le
cas du portfolio d’erreurs. The Canadian Journal of
Applied Linguistics. 16,1: 111-134.
Ellis, R. (2008). The study of second language acquisition.
Oxford, UK : Oxford University Press.
Harper, S.R. and Quaye, S.J. (2009). Student Engagement
in Higher Education. New York and London:
Routledge.
Kuh, G. (2009), Organizational culture and student
persistence: Prospects and puzzles. Journal of College
Student Retention, 3(1) 23-9.
Lawson, R., Taylor, T., Herbert, J., Fallshaw, E., French,
E., Hall, C., Kinash, S., and Summers, J. (2012, July).
Strategies to Engage Academics in Assuring Graduate
Attributes. Paper presented at the Eighth International
Conference on Education, Samos, Greece. Retrieved
on June 5, 2018, from http://
eprints.qut.edu.au/52630/2/52630.pdf
Lyster, R. et Ranta, L. (1997). Corrective feedback and
learner uptake: Negotiation of form in communicative
classrooms. Studies in Second Language Acquisition,
19, 37-66.
Popham. J. (2008). Transformative Assessment.
Alexandria: VA: Association for Supervision and
Curriculum Development.
Schmidt, R. (1990). The role of consciousness in second
language learning. Applied
Linguistics, 11, 129-158.
Zaiane, O.R. and Yacef, K. (2015). MOOCs are not
MOOCs Yet: Requirements for a True MOOC or
MOOC 2.0. Educational Letter, 11 (2), 17 21.
A2E 2019 - Special Session on Analytics in Educational Environments
680